Multi-species Bayesian hierarchical mark-recapture modelling across elevational gradients

These datasets and code underpin the analyses presented in the study by Neate-Clegg et al., (minor revisions, <i>Ornithological Applications</i>). The files comprise two text input files, two csv input files, and 11 R scripts. Together, these files can be used to run three multi-species Bayesian hierarchical mark-recapture models of increasing complexity. The first model estimates average demographic rates (capture probability, residence probability, apparent survival probability, recruitment rate, and realised population growth rate) for each species. The second model estimates how demographic rates vary across elevation (five sites). The third model estimates how demographic rates vary over both elevation and time (year).The two text files provide capture histories for differing subsets of species based on sample size: "All Species.txt" for the first model (17 species) and "Most Species.txt" for the second and third models (6 species).For the two csv files, "Capture rates.csv" provides species-level meta-data; "Effort.csv" provides meta-data on survey effort (number of banding days per year).Of the 11 R scripts, five ("data-capture_history.R", "marray.R", "mcmc.R", "nimble-wrapper.R", and "tools.R") are almost unchanged from the original writers, Telenský et al., 2023, <i>Methods in Ecology and Evolution.</i> These scripts mostly provide functions and wrappers to set up the code and run the Bayesian models in NIMBLE.The three models are split across six R scripts. For each model (indexed from 1 to 3), there is a script containing the name "Telensky" for the model setup and output, and a separate NIMBLE script (.nimble.R) for the model code itself (lacking the name "Telensky"). For example, to run the first model, one could open the file "1. Telensky Base Model Annotated.R" and run the code and it will use the model script contained within "1. Base Model.nimble.R" to run the model. All of these scripts are adapted from Telenský et al., 2023, to be multi-species, multi-site models.<br>